8th Annual Conference of the International Speech Communication Association

Antwerp, Belgium
August 27-31, 2007

The Relevance of Feature Type for the Automatic Classification of Emotional User States: Low Level Descriptors and Functionals

Björn Schuller (1), Anton Batliner (2), Dino Seppi (3), Stefan Steidl (2), Thurid Vogt (4), Johannes Wagner (4), Laurence Devillers (5), Laurence Vidrascu (5), Noam Amir (6), Loic Kessous (6), Vered Aharonson (7)

(1) Technische Universität München, Germany
(2) Friedrich-Alexander-Universität Erlangen-Nürnberg, Germany
(3) FBK-irst, Italy
(4) University of Augsburg, Germany
(5) LIMSI, France
(6) Tel Aviv University, Israel
(7) Tel Aviv Academic College of Engineering, Israel

In this paper, we report on classification results for emotional user states (4 classes, German database of children interacting with a pet robot). Six sites computed acoustic and linguistic features independently from each other, following in part different strategies. A total of 4244 features were pooled together and grouped into 12 low level descriptor types and 6 functional types. For each of these groups, classification results using Support Vector Machines and Random Forests are reported for the full set of features, and for 150 features each with the highest individual Information Gain Ratio. The performance for the different groups varies mostly between ≈ 50% and ≈ 60%.

Full Paper

Bibliographic reference.  Schuller, Björn / Batliner, Anton / Seppi, Dino / Steidl, Stefan / Vogt, Thurid / Wagner, Johannes / Devillers, Laurence / Vidrascu, Laurence / Amir, Noam / Kessous, Loic / Aharonson, Vered (2007): "The relevance of feature type for the automatic classification of emotional user states: low level descriptors and functionals", In INTERSPEECH-2007, 2253-2256.